An unauthenticated remote information disclosure vulnerability exists in the quantization engine component of the Ollama software, allowing an attacker to read and exfiltrate data from the server's heap memory. This disclosure could expose sensitive information, enable further exploitation, and facilitate stealthy persistence within the system.

The affected product is Ollama AI: Ollama. Specific version details are not provided in the current advisory, so all deployed instances of Ollama should be considered potentially impacted until an official update is issued.

The vulnerability permits unauthorized remote attackers to directly read server memory without authentication, a high-impact vector that may lead to data compromise. No CVSS score or EPSS metric is available, and the issue is not listed in the CISA KEV catalog, but the potential for sensitive data exposure and subsequent compromise indicates a significant risk. The likely attack vector is through the network-facing quantization engine exposed by Ollama, which can be accessed by unauthenticated users.